diff --git a/khmer/_cpy_smallcountgraph.hh b/khmer/_cpy_smallcountgraph.hh
index 40c18f339c..acfa2f5551 100644
--- a/khmer/_cpy_smallcountgraph.hh
+++ b/khmer/_cpy_smallcountgraph.hh
@@ -46,40 +46,7 @@ typedef struct {
 
 static void khmer_smallcountgraph_dealloc(khmer_KSmallCountgraph_Object * obj);
 
-static
-PyObject *
-smallcount_get_raw_tables(khmer_KSmallCountgraph_Object * self, PyObject * args)
-{
-    SmallCountgraph * countgraph = self->countgraph;
-
-    khmer::Byte ** table_ptrs = countgraph->get_raw_tables();
-    std::vector<uint64_t> sizes = countgraph->get_tablesizes();
-
-    PyObject * raw_tables = PyList_New(sizes.size());
-    for (unsigned int i=0; i<sizes.size(); ++i) {
-        Py_buffer buffer;
-        int res = PyBuffer_FillInfo(&buffer, NULL, table_ptrs[i],
-                                    sizes[i] / 2 +1, 0,
-                                    PyBUF_FULL_RO);
-        if (res == -1) {
-            return NULL;
-        }
-        PyObject * buf = PyMemoryView_FromBuffer(&buffer);
-        if(!PyMemoryView_Check(buf)) {
-            return NULL;
-        }
-        PyList_SET_ITEM(raw_tables, i, buf);
-    }
-
-    return raw_tables;
-}
-
 static PyMethodDef khmer_smallcountgraph_methods[] = {
-    {
-        "get_raw_tables",
-        (PyCFunction)smallcount_get_raw_tables, METH_VARARGS,
-        "Get a list of the raw storage tables as memoryview objects."
-    },
     {NULL, NULL, 0, NULL}           /* sentinel */
 };
 
diff --git a/khmer/_khmer.cc b/khmer/_khmer.cc
index 47b88efead..94752069c2 100644
--- a/khmer/_khmer.cc
+++ b/khmer/_khmer.cc
@@ -2585,32 +2585,6 @@ nodegraph_update(khmer_KNodegraph_Object * me, PyObject * args)
     Py_RETURN_NONE;
 }
 
-static
-PyObject *
-nodegraph_get_raw_tables(khmer_KNodegraph_Object * self, PyObject * args)
-{
-    Nodegraph * countgraph = self->nodegraph;
-
-    khmer::Byte ** table_ptrs = countgraph->get_raw_tables();
-    std::vector<uint64_t> sizes = countgraph->get_tablesizes();
-
-    PyObject * raw_tables = PyList_New(sizes.size());
-    for (unsigned int i=0; i<sizes.size(); ++i) {
-        Py_buffer buffer;
-        int res = PyBuffer_FillInfo(&buffer, NULL, table_ptrs[i], sizes[i], 0,
-                                    PyBUF_FULL_RO);
-        if (res == -1) {
-            return NULL;
-        }
-        PyObject * buf = PyMemoryView_FromBuffer(&buffer);
-        if(!PyMemoryView_Check(buf)) {
-            return NULL;
-        }
-        PyList_SET_ITEM(raw_tables, i, buf);
-    }
-
-    return raw_tables;
-}
 
 static PyMethodDef khmer_nodegraph_methods[] = {
     {
@@ -2618,11 +2592,6 @@ static PyMethodDef khmer_nodegraph_methods[] = {
         (PyCFunction) nodegraph_update, METH_VARARGS,
         "a set update: update this nodegraph with all the entries from the other"
     },
-    {
-        "get_raw_tables",
-        (PyCFunction) nodegraph_get_raw_tables, METH_VARARGS,
-        "Get a list of the raw tables as memoryview objects"
-    },
     {NULL, NULL, 0, NULL}           /* sentinel */
 };
 
diff --git a/lib/hashgraph.hh b/lib/hashgraph.hh
index c6ebe4c9c1..ab02faa1f4 100644
--- a/lib/hashgraph.hh
+++ b/lib/hashgraph.hh
@@ -267,6 +267,12 @@ class Countgraph : public khmer::Hashgraph
 public:
     explicit Countgraph(WordLength ksize, std::vector<uint64_t> sizes)
         : Hashgraph(ksize, new ByteStorage(sizes)) { } ;
+
+    // get access to raw tables.
+    Byte ** get_raw_tables()
+    {
+        return ((ByteStorage*)store)->get_raw_tables();
+    }
 };
 
 // Hashgraph-derived class with NibbleStorage.
diff --git a/lib/hashtable.hh b/lib/hashtable.hh
index e5cc4a1d5e..e35b6f6df6 100644
--- a/lib/hashtable.hh
+++ b/lib/hashtable.hh
@@ -277,12 +277,6 @@ public:
     void get_kmer_counts(const std::string &s,
                          std::vector<BoundedCounterType> &counts) const;
 
-    // get access to raw tables.
-    Byte ** get_raw_tables()
-    {
-        return store->get_raw_tables();
-    }
-
     // find the minimum k-mer count in the given sequence
     BoundedCounterType get_min_count(const std::string &s);
 
diff --git a/lib/khmer.hh b/lib/khmer.hh
index 0dcefe3bd3..943fb7eb3a 100644
--- a/lib/khmer.hh
+++ b/lib/khmer.hh
@@ -69,6 +69,7 @@ private:\
     className(const className&);\
     const className& operator=(const className&)
 
+#include <atomic>
 #include <set>
 #include <map>
 #include <queue>
@@ -121,6 +122,8 @@ typedef unsigned short int BoundedCounterType;
 
 // A single-byte type.
 typedef unsigned char Byte;
+using AtomicByte = std::atomic<uint8_t>;
+
 
 typedef void (*CallbackFn)(const char * info, void * callback_data,
                            unsigned long long n_reads,
diff --git a/lib/storage.cc b/lib/storage.cc
index 0523796980..5fdc9cbe6a 100644
--- a/lib/storage.cc
+++ b/lib/storage.cc
@@ -872,7 +872,7 @@ void NibbleStorage::load(std::string infilename, WordLength& ksize)
         _n_tables = (unsigned int) save_n_tables;
         _occupied_bins = save_occupied_bins;
 
-        _counts = new Byte*[_n_tables];
+        _counts = new AtomicByte*[_n_tables];
         for (unsigned int i = 0; i < _n_tables; i++) {
             _counts[i] = NULL;
         }
@@ -887,7 +887,7 @@ void NibbleStorage::load(std::string infilename, WordLength& ksize)
             tablesize = save_tablesize;
             _tablesizes.push_back(tablesize);
 
-            _counts[i] = new Byte[tablebytes];
+            _counts[i] = new AtomicByte[tablebytes];
 
             unsigned long long loaded = 0;
             while (loaded != tablebytes) {
diff --git a/lib/storage.hh b/lib/storage.hh
index 1a1bc4787b..6a67f96a38 100644
--- a/lib/storage.hh
+++ b/lib/storage.hh
@@ -38,10 +38,6 @@ Contact: khmer-project@idyll.org
 #ifndef STORAGE_HH
 #define STORAGE_HH
 
-#include <cassert>
-#include <array>
-#include <mutex>
-using MuxGuard = std::lock_guard<std::mutex>;
 
 namespace khmer
 {
@@ -67,7 +63,6 @@ public:
     virtual BoundedCounterType test_and_set_bits( HashIntoType khash ) = 0;
     virtual void add(HashIntoType khash) = 0;
     virtual const BoundedCounterType get_count(HashIntoType khash) const = 0;
-    virtual Byte ** get_raw_tables() = 0;
 
     void set_use_bigcount(bool b);
     bool get_use_bigcount();
@@ -214,13 +209,6 @@ public:
         return 1;
     }
 
-    // Writing to the tables outside of defined methods has undefined behavior!
-    // As such, this should only be used to return read-only interfaces
-    Byte ** get_raw_tables()
-    {
-        return _counts;
-    }
-
     void update_from(const BitStorage&);
 };
 
@@ -246,9 +234,8 @@ protected:
     size_t _n_tables;
     uint64_t _occupied_bins;
     uint64_t _n_unique_kmers;
-    std::array<std::mutex, 32> mutexes;
     static constexpr uint8_t _max_count{15};
-    Byte ** _counts;
+    AtomicByte ** _counts;
 
     // Compute index into the table, this retrieves the correct byte
     // which you then need to select the correct nibble from
@@ -271,8 +258,6 @@ public:
     NibbleStorage(std::vector<uint64_t>& tablesizes) :
         _tablesizes{tablesizes}, _occupied_bins{0}, _n_unique_kmers{0}
     {
-        // to allow more than 32 tables increase the size of mutex pool
-        assert(_n_tables <= 32);
         _allocate_counters();
     }
 
@@ -293,13 +278,13 @@ public:
     {
         _n_tables = _tablesizes.size();
 
-        _counts = new Byte*[_n_tables];
+        _counts = new AtomicByte*[_n_tables];
 
         for (size_t i = 0; i < _n_tables; i++) {
             const uint64_t tablesize = _tablesizes[i];
             const uint64_t tablebytes = tablesize / 2 + 1;
 
-            _counts[i] = new Byte[tablebytes];
+            _counts[i] = new AtomicByte[tablebytes];
             memset(_counts[i], 0, tablebytes);
         }
     }
@@ -317,12 +302,12 @@ public:
         bool is_new_kmer = false;
 
         for (unsigned int i = 0; i < _n_tables; i++) {
-            MuxGuard g(mutexes[i]);
-            Byte* const table(_counts[i]);
+            AtomicByte* const table(_counts[i]);
             const uint64_t idx = _table_index(khash, _tablesizes[i]);
             const uint8_t mask = _mask(khash, _tablesizes[i]);
             const uint8_t shift = _shift(khash, _tablesizes[i]);
-            const uint8_t current_count = (table[idx] & mask) >> shift;
+            uint8_t current_tbl = table[idx];
+            uint8_t current_count = (current_tbl & mask) >> shift;
 
             if (!is_new_kmer) {
                 if (current_count == 0) {
@@ -342,8 +327,25 @@ public:
             }
 
             // increase count, no checking for overflow
-            const uint8_t new_count = (current_count + 1) << shift;
-            table[idx] = (table[idx] & ~mask) | (new_count & mask);
+            // current_tbl and new_tbl are the current and new bit packed values
+            // for the idx'th byte of the table.
+            // compare_exchange_weak will update the value of table[idx] if
+            // current_tbl is the current value (hasn't been changed by a
+            // different thread) if they differ the value actually stored
+            // in table[idx] is written to current_tbl so this is a
+            // compare-and-swap loop
+            uint8_t new_count = (current_count + 1) << shift;
+            uint8_t new_tbl = (current_tbl & ~mask) | (new_count & mask);
+
+            while(!table[idx].compare_exchange_weak(current_tbl, new_tbl)) {
+                current_count = (current_tbl & mask) >> shift;
+                new_count = (current_count + 1);
+                if (new_count > _max_count) {
+                  break;
+                }
+                new_count <<= shift;
+                new_tbl = (current_tbl & ~mask) | (new_count & mask);
+            }
         }
 
         if (is_new_kmer) {
@@ -358,7 +360,7 @@ public:
 
         // get the minimum count across all tables
         for (unsigned int i = 0; i < _n_tables; i++) {
-            const Byte* table(_counts[i]);
+            const AtomicByte* table(_counts[i]);
             const uint64_t idx = _table_index(khash, _tablesizes[i]);
             const uint8_t mask = _mask(khash, _tablesizes[i]);
             const uint8_t shift = _shift(khash, _tablesizes[i]);
@@ -391,10 +393,6 @@ public:
     void save(std::string outfilename, WordLength ksize);
     void load(std::string infilename, WordLength& ksize);
 
-    Byte ** get_raw_tables()
-    {
-        return _counts;
-    }
 };
 
 
diff --git a/tests/test_countgraph.py b/tests/test_countgraph.py
index 87a0dee1db..a73f306b04 100644
--- a/tests/test_countgraph.py
+++ b/tests/test_countgraph.py
@@ -209,14 +209,11 @@ def test_get_raw_tables():
 
 
 def test_get_raw_tables_smallcountgraph():
-    # for the same number of entries a SmallCountgraph uses ~half the memory
-    # of a normal Countgraph
+    # smallcountgraphs store individual counts packed into a byte, the raw
+    # tables probably do not give users what they expect (something that can be
+    # given to numpy.frombuffer)
     ht = khmer.SmallCountgraph(20, 1e5, 4)
-    tables = ht.get_raw_tables()
-
-    for size, table in zip(ht.hashsizes(), tables):
-        assert isinstance(table, memoryview)
-        assert size // 2 + 1 == len(table)
+    assert not hasattr(ht, 'get_raw_tables')
 
 
 def test_get_raw_tables_view():
@@ -229,18 +226,6 @@ def test_get_raw_tables_view():
         assert sum(tab.tolist()) == 1
 
 
-def test_get_raw_tables_view_smallcountgraph():
-    ht = khmer.SmallCountgraph(4, 1e5, 4)
-    tables = ht.get_raw_tables()
-    for tab in tables:
-        assert sum(tab.tolist()) == 0
-    ht.consume('AAAA')
-    # the actual count is 1 but stored in the first 4bits of a Byte
-    # and so becomes 16
-    for tab in tables:
-        assert sum(tab.tolist()) == int('00010000', 2)
-
-
 @pytest.mark.huge
 def test_toobig():
     try:
diff --git a/tests/test_nodegraph.py b/tests/test_nodegraph.py
index 78a3fef410..b27acf8f5f 100644
--- a/tests/test_nodegraph.py
+++ b/tests/test_nodegraph.py
@@ -553,15 +553,11 @@ def test_extract_unique_paths_4():
 
 
 def test_get_raw_tables():
+    # nodegraphs store individual bits packed into a byte, the raw tables
+    # probably do not give users what they expect (something that can be
+    # given to numpy.frombuffer)
     kh = khmer.Nodegraph(10, 1e6, 4)
-    kh.consume('ATGGAGAGAC')
-    kh.consume('AGTGGCGATG')
-    kh.consume('ATAGACAGGA')
-    tables = kh.get_raw_tables()
-
-    for size, table in zip(kh.hashsizes(), tables):
-        assert isinstance(table, memoryview)
-        assert size == len(table)
+    assert not hasattr(kh, 'get_raw_tables')
 
 
 def test_simple_median():